Direct answer: when Ulcer Index changes its “feel”
Ulcer Index does not behave differently because of a single market “condition” like bull versus bear. Instead, it behaves differently when the shape of downside moves within the measurement window changes—especially the depth and persistence of drawdowns, and the way prices are sampled.
In practice, Ulcer Index tends to look different across these market situations:
- Deep, infrequent sell-offs vs shallow, frequent dips: deeper drawdowns generally affect the index more.
- Short spikes vs prolonged drawdown periods: sustained declines can raise the index because many observations remain below recent highs.
- Smooth trends vs choppy mean-reverting movement: choppy recoveries can keep drawdowns smaller, while steady drift lower can keep drawdowns elevated.
- Higher vs lower data granularity (sampling frequency): different bar or tick frequencies can change how often new “local highs” are reached and how drawdowns are counted.
These are conditional effects in how the calculation represents downside risk; they are not forecasts.
Mechanics: what Ulcer Index actually measures
Ulcer Index is a downside-focused measure derived from drawdowns over a selected window. Conceptually:
- Identify the running peak (a “recent high”) within the window.
- Compute the drawdown at each time step as the percentage fall from that running peak.
- Aggregate those drawdowns (the construction penalizes larger drawdowns more strongly) to produce a single number for the window.
Because the index is built from drawdowns, it is sensitive to how far and how long the price stays below its peaks.
Stable part vs variable part:
- Stable mechanics: Ulcer Index always depends on drawdown paths relative to running highs inside the chosen window.
- Variable conditions: your results depend on what the market does inside that window and on your data choices (window length and sampling frequency).
Evidence-style comparison: two drawdown “regimes”
Option A: Deep but brief drawdown
Assume a market drops significantly from a local peak and then quickly recovers to a new peak within the same window. The drawdown path contains a period of large drawdown values, followed by smaller or resetting values after recovery.
Expected conditional behavior: the index typically reflects the importance of that depth more than the length alone, because the drawdown magnitude dominates the aggregation.
Option B: Shallow but persistent drawdown
Now assume instead that the market drifts downward more slowly, with many consecutive observations showing moderate drawdowns before eventually recovering.
Expected conditional behavior: the index can be higher than in Option A even if the maximum drawdown is not as large, because the calculation accumulates downside across many time steps.
Key point
Both conditions produce different drawdown shapes, and Ulcer Index responds to drawdown shape. The comparison shows why “market condition” is really shorthand for drawdown depth, duration, and path characteristics.
Limitations and risks: where readings can fail
Even with correct mechanics, several limitations can make Ulcer Index comparisons unreliable:
- Window and sampling dependence: Changing the measurement window or using different time frequencies can alter drawdowns and running highs, which changes the index.
- Data handling and corporate actions: Splits, dividends, missing data, or inconsistent price adjustment can distort peaks and drawdowns, affecting the result.
- Regime shifts: Historical relationships between Ulcer Index and future outcomes are not guaranteed; a market can move to a different volatility or drawdown regime.
- Non-tradability of the number: Ulcer Index summarizes past drawdown behavior; it is not a standalone signal that implies future direction, timing, or safety.
Because of these issues, outcomes vary with costs, execution, and jurisdiction, and historical relationships do not establish future results.
Verification and next question
You can verify the conditional behavior by running the same calculation on two datasets that differ only in the drawdown path shape (for example, one with a deep short decline and one with a shallow long decline), while keeping the window length and sampling frequency fixed. If you see different Ulcer Index outputs, that supports the mechanism-based explanation.
If you want a deeper check, the next question to ask is: how does timeframe affect Ulcer Index, since timeframe changes what counts as drawdown events and how quickly running highs update.